import pandas as pd
pd.set_option('display.precision', 2)
pd.set_option('display.max_columns', None)
pd.set_option('display.width', None)

df = pd.read_csv('CML_2024-09-09#3-wheat_genome_1.csv')
df

df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 5000 entries, 0 to 4999
Data columns (total 45 columns):
 #   Column   Non-Null Count  Dtype  
---  ------   --------------  -----  
 0   ZYTI     5000 non-null   float64
 1   WIBD     4999 non-null   object 
 2   KVJI     5000 non-null   object 
 3   CSTT     5000 non-null   object 
 4   OFSQ     4998 non-null   object 
 5   RXUY     4999 non-null   float64
 6   GWEE     5000 non-null   float64
 7   NANX     4999 non-null   float64
 8   PMLC     4998 non-null   float64
 9   JIEK     5000 non-null   object 
 10  GXSO     4999 non-null   float64
 11  PTAX     5000 non-null   float64
 12  RLCO     5000 non-null   object 
 13  RNOZ     5000 non-null   object 
 14  QRWO     5000 non-null   float64
 15  PSSJ     5000 non-null   object 
 16  WBYY     4999 non-null   object 
 17  BDGK     5000 non-null   float64
 18  FWKW     4999 non-null   float64
 19  PFKF     5000 non-null   float64
 20  ZKOY     4998 non-null   float64
 21  XBUG     5000 non-null   float64
 22  OMXX     5000 non-null   float64
 23  IXKB     4999 non-null   float64
 24  YKJO     4999 non-null   float64
 25  GQUZ     4999 non-null   float64
 26  VKRW     4999 non-null   float64
 27  DDTT     4999 non-null   float64
 28  QXYM     5000 non-null   float64
 29  MHEC     4997 non-null   float64
 30  TUXT     4999 non-null   object 
 31  UZSD     4999 non-null   float64
 32  AOQU     5000 non-null   float64
 33  RJWA     5000 non-null   float64
 34  VTQI     5000 non-null   float64
 35  ROQE     4998 non-null   object 
 36  ZMHO     5000 non-null   object 
 37  RTDN     5000 non-null   object 
 38  MJXP     5000 non-null   object 
 39  XYKI     4998 non-null   float64
 40  RESP__0  4999 non-null   object 
 41  RESP__1  5000 non-null   object 
 42  RESP__2  4999 non-null   float64
 43  RESP__3  4999 non-null   float64
 44  RESP__4  4999 non-null   float64
dtypes: float64(29), object(16)
memory usage: 1.7+ MB

for col in df.columns:
    if df[col].dtype == 'object':
        print(df[col].apply(type).value_counts())

WIBD
<class 'str'>      4999
<class 'float'>       1
Name: count, dtype: int64
KVJI
<class 'str'>    5000
Name: count, dtype: int64
CSTT
<class 'str'>    5000
Name: count, dtype: int64
OFSQ
<class 'str'>      4998
<class 'float'>       2
Name: count, dtype: int64
JIEK
<class 'str'>    5000
Name: count, dtype: int64
RLCO
<class 'str'>    5000
Name: count, dtype: int64
RNOZ
<class 'str'>    5000
Name: count, dtype: int64
PSSJ
<class 'str'>    5000
Name: count, dtype: int64
WBYY
<class 'str'>      4999
<class 'float'>       1
Name: count, dtype: int64
TUXT
<class 'str'>      4999
<class 'float'>       1
Name: count, dtype: int64
ROQE
<class 'str'>      4998
<class 'float'>       2
Name: count, dtype: int64
ZMHO
<class 'str'>    5000
Name: count, dtype: int64
RTDN
<class 'str'>    5000
Name: count, dtype: int64
MJXP
<class 'str'>    5000
Name: count, dtype: int64
RESP__0
<class 'str'>      4999
<class 'float'>       1
Name: count, dtype: int64
RESP__1
<class 'str'>    5000
Name: count, dtype: int64

for col in df.columns:
    if df[col].dtype == 'object':
        val_types = df[col].apply(type)
        if val_types.nunique() > 1:
            float_indices = df.index[df[col].apply(type) == float].tolist()
            null_indices = df.index[df[col].isnull()].tolist()
            floats_same_nulls = float_indices == null_indices
            print(f"{col}: {floats_same_nulls}, {float_indices} floats, {null_indices} nulls")

WIBD: True, [1032] floats, [1032] nulls
OFSQ: True, [191, 2187] floats, [191, 2187] nulls
WBYY: True, [215] floats, [215] nulls
TUXT: True, [3138] floats, [3138] nulls
ROQE: True, [1738, 2247] floats, [1738, 2247] nulls
RESP__0: True, [1644] floats, [1644] nulls

def safe_float(x):
    try:
        return float(x)
    except ValueError:
        return x  
    
for col in df.columns:
    if df[col].dtype == 'object':
        col_float = df[col].apply(lambda x: safe_float(x) if not pd.isna(x) else x)
        non_float_mask = ~col_float.apply(lambda x: isinstance(x, float) or pd.isna(x))
        
        if non_float_mask.any():
            print(f"\nColumn: {col}")
            non_float_vals = df[col][non_float_mask].index
            print(col_float[non_float_mask].tolist())

Column: WIBD
['<Err>']

Column: KVJI
['<Err>']

Column: CSTT
['<Err>', '<Err>']

Column: OFSQ
['<Err>']

Column: JIEK
['<Err>']

Column: RLCO
['<Err>']

Column: RNOZ
['<Err>']

Column: PSSJ
['<Err>']

Column: WBYY
['<Err>']

Column: TUXT
['<Err>']

Column: ROQE
['<Err>']

Column: ZMHO
['<Err>']

Column: RTDN
['<Err>', '<Err>']

Column: MJXP
['<Err>', '<Err>']

Column: RESP__0
['<Err>']

Column: RESP__1
['<Err>', '<Err>']

import missingno as msno
msno.matrix(df)

<Axes: >

msno.heatmap(df)

<Axes: >

df_no_nan = df.dropna()
print(df_no_nan.shape)

(4970, 45)

df_err_mask = (df == '<Err>').any(axis=1)
df_no_err = df.drop(df[df_err_mask].index)
print(df_no_err.shape)

(4980, 45)

df_no_err_nan = df_no_err.dropna()
print(df_no_err_nan.shape)

(4950, 45)

import numpy as np
df_resp_na_mask = df_no_err.iloc[:, -5:].isna().any(axis=1)
print(df_resp_na_mask.sum())
df_no_err_resp = df_no_err.drop(df_no_err[df_resp_na_mask].index)
print(df_no_err_resp.shape)

4
(4976, 45)

from decimal import Decimal

def safe_decimal(x):
    try:
        return Decimal(x)
    except (ValueError, TypeError):
        return x

for col in df_no_err_resp.columns:
    if df_no_err_resp[col].dtype == 'object':
        # cast strings to float and decimal
        col_float = df_no_err_resp[col].dropna().apply(safe_float)
        col_decimal = df_no_err_resp[col].dropna().apply(safe_decimal)

        # cast float to decimal
        col_float = col_float.apply(Decimal)
        
        float_decimal_diff = col_float != col_decimal
        if float_decimal_diff.any():
            diff_indices = float_decimal_diff[float_decimal_diff].index.tolist()
            float_diffs = col_float.loc[diff_indices]
            decimal_diffs = col_decimal.loc[diff_indices]
            print(f"Float diffs in {col}")
            max_diff = Decimal(0)
            max_diff_index = 0
            for i in range(len(float_diffs)):
                diff = abs(float_diffs.iloc[i] - decimal_diffs.iloc[i])
                if diff > max_diff:
                    max_diff = diff
                    max_diff_index = i

            print(f"Max diff: {max_diff} where {float_diffs.iloc[max_diff_index]} != {decimal_diffs.iloc[max_diff_index]}")

Float diffs in WIBD
Max diff: 0.0000046826171875 where 109156811357.5259246826171875 != 109156811357.52592
Float diffs in KVJI
Max diff: 0.0000046875 where -118098088887.4189453125 != -118098088887.41895
Float diffs in CSTT
Max diff: 0.00001455078125 where -162216975264.58758544921875 != -162216975264.5876
Float diffs in OFSQ
Max diff: 0.0000021875 where 142949236984.4892578125 != 142949236984.48926
Float diffs in JIEK
Max diff: 0.000002763671875 where 178946755275.296417236328125 != 178946755275.29642
Float diffs in RLCO
Max diff: 0.000001435546875 where 161853626366.520721435546875 != 161853626366.52072
Float diffs in RNOZ
Max diff: 0.000002275390625 where 126073385674.965667724609375 != 126073385674.96567
Float diffs in PSSJ
Max diff: 0.0000142578125 where -197017749143.1671142578125 != -197017749143.1671
Float diffs in WBYY
Max diff: 0.000004326171875 where 185202097981.529815673828125 != 185202097981.52982
Float diffs in TUXT
Max diff: 0.00001240234375 where 152953352189.59478759765625 != 152953352189.5948
Float diffs in ROQE
Max diff: 0.00000478515625 where -194479254725.15240478515625 != -194479254725.1524
Float diffs in ZMHO
Max diff: 0.0000046044921875 where 100697248396.2063446044921875 != 100697248396.20634
Float diffs in RTDN
Max diff: 8.7890625E-8 where -144693497917.132659912109375 != -144693497917.13266
Float diffs in MJXP
Max diff: 9.765625E-8 where -162831873525.41510009765625 != -162831873525.4151
Float diffs in RESP__0
Max diff: 0.00000380859375 where 36856196793.27984619140625 != 36856196793.27985
Float diffs in RESP__1
Max diff: 0.0000134765625 where 192457552679.2161865234375 != 192457552679.2162

Decimal(0.0000046826171875)/Decimal(109156811357.5259246826171875) * 100

Decimal('4.289807598137715388151535604E-15')

df_float = df_no_err_resp.copy()
for col in df_float.columns:
    if df_float[col].dtype == 'object':
        df_float[col] = df_float[col].apply(lambda x: safe_float(x) if not pd.isna(x) else x)

df_float.info()

<class 'pandas.core.frame.DataFrame'>
Index: 4976 entries, 0 to 4999
Data columns (total 45 columns):
 #   Column   Non-Null Count  Dtype  
---  ------   --------------  -----  
 0   ZYTI     4976 non-null   float64
 1   WIBD     4975 non-null   float64
 2   KVJI     4976 non-null   float64
 3   CSTT     4976 non-null   float64
 4   OFSQ     4974 non-null   float64
 5   RXUY     4975 non-null   float64
 6   GWEE     4976 non-null   float64
 7   NANX     4975 non-null   float64
 8   PMLC     4974 non-null   float64
 9   JIEK     4976 non-null   float64
 10  GXSO     4975 non-null   float64
 11  PTAX     4976 non-null   float64
 12  RLCO     4976 non-null   float64
 13  RNOZ     4976 non-null   float64
 14  QRWO     4976 non-null   float64
 15  PSSJ     4976 non-null   float64
 16  WBYY     4975 non-null   float64
 17  BDGK     4976 non-null   float64
 18  FWKW     4975 non-null   float64
 19  PFKF     4976 non-null   float64
 20  ZKOY     4974 non-null   float64
 21  XBUG     4976 non-null   float64
 22  OMXX     4976 non-null   float64
 23  IXKB     4975 non-null   float64
 24  YKJO     4975 non-null   float64
 25  GQUZ     4975 non-null   float64
 26  VKRW     4975 non-null   float64
 27  DDTT     4975 non-null   float64
 28  QXYM     4976 non-null   float64
 29  MHEC     4973 non-null   float64
 30  TUXT     4975 non-null   float64
 31  UZSD     4975 non-null   float64
 32  AOQU     4976 non-null   float64
 33  RJWA     4976 non-null   float64
 34  VTQI     4976 non-null   float64
 35  ROQE     4974 non-null   float64
 36  ZMHO     4976 non-null   float64
 37  RTDN     4976 non-null   float64
 38  MJXP     4976 non-null   float64
 39  XYKI     4974 non-null   float64
 40  RESP__0  4976 non-null   float64
 41  RESP__1  4976 non-null   float64
 42  RESP__2  4976 non-null   float64
 43  RESP__3  4976 non-null   float64
 44  RESP__4  4976 non-null   float64
dtypes: float64(45)
memory usage: 1.7 MB

import seaborn as sns
import matplotlib.pyplot as plt

cols_to_plot = df_float.columns[df_float.isna().any()].to_list()
fig, axes = plt.subplots(nrows=len(cols_to_plot), ncols=2, figsize=(12, 50))
for i, col in enumerate(cols_to_plot):
    data = df_float[col].dropna()

    # kde plot
    sns.kdeplot(data, ax=axes[i][0], fill=True)
    axes[i][0].set_title(f'Distribution of {col}')
    axes[i][0].set_xlabel('')
    axes[i][0].set_ylabel('')
    axes[i][0].set_xlim(data.min(), data.max())

    # log plot
    sns.kdeplot(data, ax=axes[i][1], fill=True)
    axes[i][1].set_title(f'Distribution of log {col}')
    axes[i][1].set_xlabel('')
    axes[i][1].set_ylabel('')
    axes[i][1].set_xlim(data.min(), data.max())
    axes[i][1].set_xscale('log')

plt.tight_layout()
plt.show()

df_impute = df_float.copy()
for col in df_impute.columns:
    if df_impute[col].dtype == 'float64':
        median_val = df_impute[col].median()
        df_impute[col].fillna(median_val, inplace=True)

df_impute.info()

<class 'pandas.core.frame.DataFrame'>
Index: 4976 entries, 0 to 4999
Data columns (total 45 columns):
 #   Column   Non-Null Count  Dtype  
---  ------   --------------  -----  
 0   ZYTI     4976 non-null   float64
 1   WIBD     4976 non-null   float64
 2   KVJI     4976 non-null   float64
 3   CSTT     4976 non-null   float64
 4   OFSQ     4976 non-null   float64
 5   RXUY     4976 non-null   float64
 6   GWEE     4976 non-null   float64
 7   NANX     4976 non-null   float64
 8   PMLC     4976 non-null   float64
 9   JIEK     4976 non-null   float64
 10  GXSO     4976 non-null   float64
 11  PTAX     4976 non-null   float64
 12  RLCO     4976 non-null   float64
 13  RNOZ     4976 non-null   float64
 14  QRWO     4976 non-null   float64
 15  PSSJ     4976 non-null   float64
 16  WBYY     4976 non-null   float64
 17  BDGK     4976 non-null   float64
 18  FWKW     4976 non-null   float64
 19  PFKF     4976 non-null   float64
 20  ZKOY     4976 non-null   float64
 21  XBUG     4976 non-null   float64
 22  OMXX     4976 non-null   float64
 23  IXKB     4976 non-null   float64
 24  YKJO     4976 non-null   float64
 25  GQUZ     4976 non-null   float64
 26  VKRW     4976 non-null   float64
 27  DDTT     4976 non-null   float64
 28  QXYM     4976 non-null   float64
 29  MHEC     4976 non-null   float64
 30  TUXT     4976 non-null   float64
 31  UZSD     4976 non-null   float64
 32  AOQU     4976 non-null   float64
 33  RJWA     4976 non-null   float64
 34  VTQI     4976 non-null   float64
 35  ROQE     4976 non-null   float64
 36  ZMHO     4976 non-null   float64
 37  RTDN     4976 non-null   float64
 38  MJXP     4976 non-null   float64
 39  XYKI     4976 non-null   float64
 40  RESP__0  4976 non-null   float64
 41  RESP__1  4976 non-null   float64
 42  RESP__2  4976 non-null   float64
 43  RESP__3  4976 non-null   float64
 44  RESP__4  4976 non-null   float64
dtypes: float64(45)
memory usage: 1.7 MB

df_impute.describe()

pd.plotting.scatter_matrix(df_impute, figsize=(10, 10), diagonal='hist')
plt.tight_layout
plt.show()

c:\Python311\Lib\site-packages\pandas\plotting\_matplotlib\misc.py:122: RuntimeWarning: invalid value encountered in cast
  if np.all(locs == locs.astype(int)):

responses = ['RESP__0', 'RESP__1', 'RESP__2', 'RESP__3', 'RESP__4']
pd.plotting.scatter_matrix(df_impute[responses], figsize=(10, 10), diagonal='hist')
plt.tight_layout
plt.show()

c:\Python311\Lib\site-packages\pandas\plotting\_matplotlib\misc.py:122: RuntimeWarning: invalid value encountered in cast
  if np.all(locs == locs.astype(int)):

plt.figure(figsize=(10,100))
sns.violinplot(data=df_impute, orient='h')
plt.show()

df_no_outliers = df_impute.copy()
no_outliers = ['QRWO', 'UZSD']

both_outliers = ['ZYTI', 'WIBD', 'OFSQ', 'RXUY', 'NANX', 'JIEK', 'GXSO', 'PTAX', 'RNOZ', 'PSSJ', 'WBYY', 'XBUG', 'IXKB', 'YKJO', 'DDTT', 'QXYM', 'MHEC', 'TUXT', 'AOQU', 'VTQI', 'ROQE', 'ZMHO', 'XYKI', 'RESP__2']
n = 5 # also k factor
for col in both_outliers:
    median = df_no_outliers[col].median()
    mad = (df_no_outliers[col] - median).abs().median()
    upper_bound = median + n * mad
    lower_bound = median - n * mad
    upper_cond = df_no_outliers[col] > upper_bound
    lower_cond = df_no_outliers[col] < lower_bound
    df_no_outliers.loc[upper_cond | lower_cond, col] = np.nan
    new_median = df_no_outliers[col].median()
    df_no_outliers[col] = df_no_outliers[col].fillna(new_median)

n = 5 # also k factor

left_outliers = ['KVJI', 'CSTT', 'GWEE', 'BDGK', 'PFKF', 'VKRW', 'RTDN', 'MJXP', 'RESP__0']
for col in left_outliers:
    median = df_no_outliers[col].median()
    mad = (df_no_outliers[col] - median).abs().median()
    lower_bound = median - n * mad
    lower_cond = df_no_outliers[col] < lower_bound
    df_no_outliers.loc[lower_cond, col] = np.nan
    new_median = df_no_outliers[col].median()
    df_no_outliers[col] = df_no_outliers[col].fillna(new_median)

right_outliers = ['PMLC', 'RLCO', 'FWKW', 'ZKOY', 'OMXX', 'GQUZ', 'RJWA', 'RESP__1', 'RESP__3', 'RESP__4']
for col in right_outliers:
    median = df_no_outliers[col].median()
    mad = (df_no_outliers[col] - median).abs().median()
    upper_bound = median + n * mad
    upper_cond = df_no_outliers[col] > upper_bound
    df_no_outliers.loc[upper_cond, col] = np.nan
    new_median = df_no_outliers[col].median()
    df_no_outliers[col] = df_no_outliers[col].fillna(new_median)

num_cols = len(no_outliers) + len(both_outliers) + len(left_outliers) + len(right_outliers)
print(num_cols) 

plt.figure(figsize=(10,90))
sns.violinplot(data=df_no_outliers.iloc[:, :-5], orient='h')
plt.show()

45

plt.figure(figsize=(10,10))
sns.violinplot(data=df_no_outliers.iloc[:, -5:], orient='h')
plt.show()

df_no_outliers.info()

<class 'pandas.core.frame.DataFrame'>
Index: 4976 entries, 0 to 4999
Data columns (total 45 columns):
 #   Column   Non-Null Count  Dtype  
---  ------   --------------  -----  
 0   ZYTI     4976 non-null   float64
 1   WIBD     4976 non-null   float64
 2   KVJI     4976 non-null   float64
 3   CSTT     4976 non-null   float64
 4   OFSQ     4976 non-null   float64
 5   RXUY     4976 non-null   float64
 6   GWEE     4976 non-null   float64
 7   NANX     4976 non-null   float64
 8   PMLC     4976 non-null   float64
 9   JIEK     4976 non-null   float64
 10  GXSO     4976 non-null   float64
 11  PTAX     4976 non-null   float64
 12  RLCO     4976 non-null   float64
 13  RNOZ     4976 non-null   float64
 14  QRWO     4976 non-null   float64
 15  PSSJ     4976 non-null   float64
 16  WBYY     4976 non-null   float64
 17  BDGK     4976 non-null   float64
 18  FWKW     4976 non-null   float64
 19  PFKF     4976 non-null   float64
 20  ZKOY     4976 non-null   float64
 21  XBUG     4976 non-null   float64
 22  OMXX     4976 non-null   float64
 23  IXKB     4976 non-null   float64
 24  YKJO     4976 non-null   float64
 25  GQUZ     4976 non-null   float64
 26  VKRW     4976 non-null   float64
 27  DDTT     4976 non-null   float64
 28  QXYM     4976 non-null   float64
 29  MHEC     4976 non-null   float64
 30  TUXT     4976 non-null   float64
 31  UZSD     4976 non-null   float64
 32  AOQU     4976 non-null   float64
 33  RJWA     4976 non-null   float64
 34  VTQI     4976 non-null   float64
 35  ROQE     4976 non-null   float64
 36  ZMHO     4976 non-null   float64
 37  RTDN     4976 non-null   float64
 38  MJXP     4976 non-null   float64
 39  XYKI     4976 non-null   float64
 40  RESP__0  4976 non-null   float64
 41  RESP__1  4976 non-null   float64
 42  RESP__2  4976 non-null   float64
 43  RESP__3  4976 non-null   float64
 44  RESP__4  4976 non-null   float64
dtypes: float64(45)
memory usage: 1.7 MB

from sklearn.decomposition import PCA

df_x = df_no_outliers.iloc[:, :40]
pca = PCA()
pca.fit(df_x)
df_x

plt.plot(range(1, len(pca.explained_variance_) + 1), pca.explained_variance_, 'bo-')
plt.xlabel('Principal Component')
plt.ylabel('Eigenvalue')
plt.title('Scree Plot')

Text(0.5, 1.0, 'Scree Plot')

plt.plot(np.cumsum(pca.explained_variance_ratio_))
plt.xlabel('Number of Components')
plt.ylabel('Cumulative Explained Variance')
plt.show()

from sklearn.preprocessing import PowerTransformer

transformer = PowerTransformer(method='yeo-johnson')
df_transformed = transformer.fit_transform(df_no_outliers)
pca_transformed = PCA()
pca_transformed.fit(df_transformed)

plt.plot(range(1, len(pca_transformed.explained_variance_) + 1), pca_transformed.explained_variance_, 'bo-')
plt.xlabel('Principal Component')
plt.ylabel('Eigenvalue')
plt.title('Scree Plot')

Text(0.5, 1.0, 'Scree Plot')

plt.plot(np.cumsum(pca_transformed.explained_variance_ratio_))
plt.xlabel('Number of Components')
plt.ylabel('Cumulative Explained Variance')
plt.show()

from sklearn.preprocessing import StandardScaler

scaler = StandardScaler()
df_scaled = scaler.fit_transform(df_no_outliers)
pca_scaler = PCA()
pca_scaler.fit(df_scaled)

plt.plot(np.cumsum(pca_scaler.explained_variance_ratio_))
plt.xlabel('Number of Components')
plt.ylabel('Cumulative Explained Variance')
plt.show()

pca = PCA(n_components=7)
np_pca = pca.fit_transform(df_x)
df_pca = pd.DataFrame(np_pca, columns=[f"PC{i}" for i in range(1, 8)])
pd.plotting.scatter_matrix(df_pca, figsize=(10, 10))

array([[<Axes: xlabel='PC1', ylabel='PC1'>,
        <Axes: xlabel='PC2', ylabel='PC1'>,
        <Axes: xlabel='PC3', ylabel='PC1'>,
        <Axes: xlabel='PC4', ylabel='PC1'>,
        <Axes: xlabel='PC5', ylabel='PC1'>,
        <Axes: xlabel='PC6', ylabel='PC1'>,
        <Axes: xlabel='PC7', ylabel='PC1'>],
       [<Axes: xlabel='PC1', ylabel='PC2'>,
        <Axes: xlabel='PC2', ylabel='PC2'>,
        <Axes: xlabel='PC3', ylabel='PC2'>,
        <Axes: xlabel='PC4', ylabel='PC2'>,
        <Axes: xlabel='PC5', ylabel='PC2'>,
        <Axes: xlabel='PC6', ylabel='PC2'>,
        <Axes: xlabel='PC7', ylabel='PC2'>],
       [<Axes: xlabel='PC1', ylabel='PC3'>,
        <Axes: xlabel='PC2', ylabel='PC3'>,
        <Axes: xlabel='PC3', ylabel='PC3'>,
        <Axes: xlabel='PC4', ylabel='PC3'>,
        <Axes: xlabel='PC5', ylabel='PC3'>,
        <Axes: xlabel='PC6', ylabel='PC3'>,
        <Axes: xlabel='PC7', ylabel='PC3'>],
       [<Axes: xlabel='PC1', ylabel='PC4'>,
        <Axes: xlabel='PC2', ylabel='PC4'>,
        <Axes: xlabel='PC3', ylabel='PC4'>,
        <Axes: xlabel='PC4', ylabel='PC4'>,
        <Axes: xlabel='PC5', ylabel='PC4'>,
        <Axes: xlabel='PC6', ylabel='PC4'>,
        <Axes: xlabel='PC7', ylabel='PC4'>],
       [<Axes: xlabel='PC1', ylabel='PC5'>,
        <Axes: xlabel='PC2', ylabel='PC5'>,
        <Axes: xlabel='PC3', ylabel='PC5'>,
        <Axes: xlabel='PC4', ylabel='PC5'>,
        <Axes: xlabel='PC5', ylabel='PC5'>,
        <Axes: xlabel='PC6', ylabel='PC5'>,
        <Axes: xlabel='PC7', ylabel='PC5'>],
       [<Axes: xlabel='PC1', ylabel='PC6'>,
        <Axes: xlabel='PC2', ylabel='PC6'>,
        <Axes: xlabel='PC3', ylabel='PC6'>,
        <Axes: xlabel='PC4', ylabel='PC6'>,
        <Axes: xlabel='PC5', ylabel='PC6'>,
        <Axes: xlabel='PC6', ylabel='PC6'>,
        <Axes: xlabel='PC7', ylabel='PC6'>],
       [<Axes: xlabel='PC1', ylabel='PC7'>,
        <Axes: xlabel='PC2', ylabel='PC7'>,
        <Axes: xlabel='PC3', ylabel='PC7'>,
        <Axes: xlabel='PC4', ylabel='PC7'>,
        <Axes: xlabel='PC5', ylabel='PC7'>,
        <Axes: xlabel='PC6', ylabel='PC7'>,
        <Axes: xlabel='PC7', ylabel='PC7'>]], dtype=object)

from sklearn.cluster import KMeans
from sklearn import metrics

def k_silhouette(X, clusters): # [11] i.e. code taken from Dr Collins - Classical Machine Learning Workbook
    K = range(2, clusters + 1)
    score = []
    for k in K:
        kmeans = KMeans(n_clusters=k)
        kmeans.fit(X)
        labels = kmeans.labels_
        score.append(metrics.silhouette_score(X, labels, metric='euclidean'))

    plt.plot(K, score, 'b*-')
    plt.xlabel('k')
    plt.ylabel('silhouette_score')

    plt.show()

k_silhouette(df_pca, 10)

sse = []
for k in range(1, 11): 
    kmeans = KMeans(n_clusters=k)
    kmeans.fit(df_pca)
    sse.append(kmeans.inertia_)
x = range(1, 11)
plt.xlabel('k')
plt.ylabel('sse')
plt.plot(x, sse, 'o-')
plt.show()

kmeans = KMeans(n_clusters=8)
cluster_predict = kmeans.fit_predict(df_pca)
df_predict = df_pca.copy()
df_predict["cluster"] = cluster_predict.astype(str)
sns.pairplot(df_predict, hue='cluster')

<seaborn.axisgrid.PairGrid at 0x2909abb1f50>

pca = PCA(n_components=7)
df_pca = pca.fit_transform(df_x)

kmeans = KMeans(n_clusters=8)
kmeans.fit(df_pca)

pca_centres = kmeans.cluster_centers_
feature_centres = pca.inverse_transform(pca_centres)
feature_centres

array([[88448001.1333175 , 61342371.38274013, 55779041.07009545,
        62915837.24315398, 50615440.57440168, 95051962.57258865,
        76637613.83933033, 72924160.82353662, 94764651.42291065,
        75646259.61444706, 52637290.81998856, 56103507.91466573,
        62889184.33427719, 53239141.20688049, 52907066.28346581,
        93036337.19079626, 58977557.15911847, 95584848.47205909,
        96476965.21871266, 86578181.19801939, 89716909.7302    ,
        88949235.23142767, 84350807.3550347 , 70087779.78973761,
        77888769.13596703, 63433933.51861309, 74531516.1879485 ,
        83140642.03988124, 86849012.42039655, 91946351.94717573,
        55318474.65979967, 58758148.1399405 , 53295452.46426091,
        56850866.29440966, 78496260.14545742, 76441699.76149927,
        65954934.78056125, 57208442.1433517 , 97212395.868873  ,
        63594843.35139872],
       [72892274.3682851 , 87122048.1358604 , 82176919.83513469,
        59002312.11712766, 77830640.34214726, 53567606.82803463,
        94746710.38687158, 59130831.19676776, 51467898.93520284,
        95518232.60773627, 60755804.33932257, 62967020.99487875,
        51726141.52262151, 83953223.29547688, 66908714.38993482,
        60558156.03383322, 72086632.72967443, 98601696.67078254,
        68873412.28426357, 90979271.34844723, 74928822.74020688,
        76302048.37738457, 93488427.17263488, 60012876.52371944,
        69616227.43053472, 76310930.5950143 , 96850540.57799242,
        89857539.45578256, 54630595.70761789, 53450788.84985921,
        89919462.11244857, 64165965.93008526, 86748319.49887954,
        73219524.23958212, 64599204.87443362, 67813076.73852381,
        65669949.3389558 , 59947377.05684397, 89962741.27646343,
        87804992.62577216],
       [68846753.95681483, 54232694.9176927 , 92490562.12401703,
        57567767.92371108, 58430472.09014502, 65513042.50014889,
        97039643.25108387, 95509298.65666305, 65101816.66010966,
        94258144.7913048 , 97733716.5426366 , 96063973.97569315,
        68439978.45228103, 92497171.29752153, 52063839.31702869,
        77511011.55095956, 54777307.27067419, 52950193.34069047,
        57886102.09435633, 54065892.02507694, 87483285.80224878,
        62182507.96444229, 65442332.30061423, 97611918.91633435,
        90244523.16785702, 84686009.88520621, 78722480.6893866 ,
        63569167.15235315, 86449853.08706391, 93517417.32740867,
        94892060.77002235, 54752473.17408752, 55099528.51087534,
        54930114.44235059, 56794860.91008878, 58167321.39506388,
        95286803.21113378, 98124097.53048937, 84504319.91864681,
        76195798.56175648],
       [70029920.43341598, 74943473.85539602, 89670532.74653591,
        68098703.08981077, 67922877.49081698, 98075491.95857327,
        52955348.06636768, 67221855.83203977, 92587208.95726386,
        57158954.3729156 , 95394123.1605093 , 90085067.44665983,
        95291141.69263053, 79564091.43609859, 76082313.67049743,
        90941329.26319003, 61117645.57269862, 59081777.78045551,
        75269291.46939188, 70641165.75759548, 95811974.99071774,
        70110448.70760664, 83256799.3220674 , 83213670.51066208,
        72982927.63536006, 55469666.57138341, 54006894.57346066,
        58168649.1661066 , 81059071.99013531, 50911568.00581528,
        66575142.99234235, 51349951.70957495, 67968333.3280002 ,
        61973427.49967094, 80793292.88465844, 86392194.1612533 ,
        50380505.7988096 , 90924717.17968261, 56208322.107401  ,
        84220670.01154138],
       [97130165.6400826 , 55034881.89096344, 69009057.4148529 ,
        94850285.4894659 , 69009497.16490617, 79136846.65249795,
        78835214.36054213, 86160086.32781935, 94215708.29696009,
        87486941.67797714, 54789728.27510196, 81537911.63291118,
        95632206.7795917 , 62038163.43922652, 59325985.90796389,
        60707173.03592721, 69840089.66438456, 97244054.27575968,
        61315784.497266  , 60909064.6186956 , 76206552.4134738 ,
        67885010.68166965, 75638406.20511276, 65858403.88780936,
        56751947.6017639 , 64585020.49812669, 64671845.19114268,
        76138847.59872685, 83362141.3093259 , 52566897.261009  ,
        99902054.09242159, 86749359.69638889, 80791653.92825657,
        50531691.77583265, 51063730.28119735, 51787039.70196834,
        73014639.77083787, 56751145.74444704, 50781841.24422266,
        63446974.21295395],
       [53549077.73088518, 97275309.36307749, 99281701.61051932,
        91250885.94151762, 83314887.46850836, 55009254.25831735,
        61259731.93381738, 62315933.17465385, 56616650.39182454,
        84057750.09629253, 95892124.31468269, 98414426.09061733,
        58206363.45371614, 51552262.08239493, 71751206.21947853,
        58353727.79446261, 66294545.98926888, 91882484.41310704,
        62066391.50483945, 53702775.03937888, 70577709.670233  ,
        56739375.35194807, 59748472.18287794, 84780786.64373136,
        70211446.6034445 , 52414916.84157054, 68308085.8161096 ,
        97409126.68734229, 76828843.87777086, 95217910.26824516,
        70029497.26391391, 64519529.79163358, 89974571.643438  ,
        93256503.70602617, 66596620.02224921, 72978381.72507808,
        79048548.07577112, 62284803.83927371, 73975014.99888073,
        72255950.08092634],
       [73136931.47101814, 67457394.7814256 , 98686778.41508463,
        58439385.88262246, 70257393.39755815, 60951887.51490064,
        84868318.8340625 , 91784722.92127025, 54772146.49581683,
        61030246.89352381, 74970352.9789898 , 82739930.24714054,
        66437536.3463517 , 86323606.06644587, 58342849.19387696,
        77535760.94803534, 94386124.08365542, 92700530.78777266,
        87048525.706248  , 56101787.38114613, 78322504.35072932,
        55485592.05458958, 84448973.45759004, 90277221.0332783 ,
        51430200.28132342, 78327801.4372856 , 95163837.31343625,
        56432413.25053175, 80123773.12356657, 78638248.19731884,
        75405902.14300266, 92731965.11416617, 65744851.5710576 ,
        71971672.62072718, 64924946.67085569, 94692465.68373637,
        62268057.22752561, 55648877.95763533, 96953832.6462976 ,
        76904601.3296504 ],
       [60400589.06806988, 52673472.49295449, 76422702.51947941,
        81090136.26367426, 75674577.53432491, 66105828.90796952,
        51531285.59157993, 80869941.14921609, 80503951.2133245 ,
        87784986.12002164, 60529332.94886982, 59553927.72815239,
        56926798.20010852, 57618252.09764703, 51936057.50121683,
        51863852.25314425, 90412958.88701965, 70123689.3075403 ,
        55855085.05607122, 95510149.58142391, 74362098.50144745,
        64393531.11228432, 58807167.26579897, 82918189.18195662,
        69684073.47647108, 84461548.61499196, 54680451.95208299,
        98569474.5337855 , 63054115.43402275, 63578437.49700574,
        61227915.81397676, 53525624.61283235, 57783218.88684861,
        62657338.02817196, 78425519.52911933, 56435790.18236001,
        84563086.31752393, 63178119.58173623, 78870998.03723364,
        82717493.00269409]])

pca_viz = PCA(n_components=2)
centres_2d = pca_viz.fit_transform(feature_centres)

plt.figure(figsize=(10, 8))
plt.scatter(centres_2d[:, 0], centres_2d[:, 1], c='red', s=200, alpha=0.8)
for i, (x, y) in enumerate(centres_2d):
    plt.annotate(f'Cluster {i}', (x, y), xytext=(5, 5), textcoords='offset points')
plt.xlabel('First Principal Component')
plt.ylabel('Second Principal Component')
plt.grid(True)
plt.show()

loadings = pca.components_.T
df_loadings = pd.DataFrame(loadings, columns=['PC1', 'PC2', 'PC3', 'PC4', 'PC5', 'PC6', 'PC7'], index=df_x.columns)
df_loadings

def get_n_features(loadings_df, n):
    n_features = {}
    for col in loadings_df.columns:
        largest_n = loadings_df[col].abs().nlargest(n)
        n_features[col] = list(zip(largest_n.index, largest_n.values))
    return n_features

n_features = get_n_features(df_loadings, 10)
for pc, features in n_features.items():
    print(f"{pc}:")
    for feature, loading in features:
        print(f"  {feature}: {loading:.4f}")

PC1:
  GXSO: 0.3976
  PTAX: 0.3212
  KVJI: 0.2956
  PFKF: 0.2862
  RTDN: 0.2620
  BDGK: 0.2497
  IXKB: 0.2324
  PMLC: 0.2310
  RNOZ: 0.2241
  XBUG: 0.1789
PC2:
  RXUY: 0.3533
  PMLC: 0.3270
  PSSJ: 0.2746
  RLCO: 0.2729
  BDGK: 0.2440
  DDTT: 0.2412
  RTDN: 0.2252
  RJWA: 0.2122
  AOQU: 0.2030
  WIBD: 0.1886
PC3:
  GWEE: 0.3723
  MJXP: 0.3505
  VKRW: 0.3442
  CSTT: 0.3271
  RNOZ: 0.2332
  FWKW: 0.2158
  OMXX: 0.1896
  DDTT: 0.1870
  GQUZ: 0.1804
  PSSJ: 0.1764
PC4:
  ROQE: 0.3825
  ZMHO: 0.3318
  WIBD: 0.3151
  JIEK: 0.3118
  FWKW: 0.2504
  GQUZ: 0.2165
  RJWA: 0.2079
  QRWO: 0.1922
  NANX: 0.1907
  OMXX: 0.1894
PC5:
  TUXT: 0.3601
  UZSD: 0.3400
  RLCO: 0.3130
  MJXP: 0.2701
  MHEC: 0.2679
  ZYTI: 0.2436
  VTQI: 0.2360
  PFKF: 0.2342
  YKJO: 0.2079
  AOQU: 0.1879
PC6:
  MHEC: 0.5237
  RNOZ: 0.3228
  PFKF: 0.3031
  XYKI: 0.2778
  QXYM: 0.2708
  BDGK: 0.2152
  RTDN: 0.1966
  CSTT: 0.1931
  UZSD: 0.1852
  OMXX: 0.1795
PC7:
  WBYY: 0.4840
  NANX: 0.2697
  WIBD: 0.2686
  GWEE: 0.2527
  YKJO: 0.2503
  JIEK: 0.2350
  XBUG: 0.2161
  GQUZ: 0.2106
  IXKB: 0.2088
  AOQU: 0.2014

df_cluster_centres = pd.DataFrame(feature_centres, columns=df_x.columns)

def get_distinguishing_features(df_cluster_centres, n):
    distinguishing_features = {}
    for i in range(len(df_cluster_centres)):
        diff_from_mean = df_cluster_centres.iloc[i] - df_cluster_centres.mean()
        top_n = diff_from_mean.abs().nlargest(n)
        relative_importance = top_n / top_n.sum()
        distinguishing_features[f'Cluster {i}'] = list(zip(top_n.index, relative_importance))
    return distinguishing_features

distinguishing_features = get_distinguishing_features(df_cluster_centres, 5)

for cluster, features in distinguishing_features.items():
    print(f"{cluster}:")
    for feature, importance in features:
        print(f"  {feature}: {importance:.4f}")

Cluster 0:
  KVJI: 0.2255
  FWKW: 0.2148
  RXUY: 0.1940
  PTAX: 0.1854
  PSSJ: 0.1803
Cluster 1:
  VKRW: 0.2182
  PMLC: 0.2071
  QXYM: 0.2036
  GWEE: 0.1860
  PFKF: 0.1851
Cluster 2:
  RTDN: 0.2341
  BDGK: 0.2279
  GXSO: 0.1838
  ZMHO: 0.1808
  GWEE: 0.1734
Cluster 3:
  RXUY: 0.2172
  RLCO: 0.2126
  JIEK: 0.1909
  BDGK: 0.1908
  RTDN: 0.1885
Cluster 4:
  MJXP: 0.2231
  RLCO: 0.2104
  ZYTI: 0.1934
  TUXT: 0.1867
  CSTT: 0.1864
Cluster 5:
  WIBD: 0.2358
  RJWA: 0.2281
  MHEC: 0.1880
  GXSO: 0.1803
  AOQU: 0.1678
Cluster 6:
  UZSD: 0.2287
  ROQE: 0.2048
  WBYY: 0.1988
  VKRW: 0.1852
  DDTT: 0.1825
Cluster 7:
  PFKF: 0.2281
  GWEE: 0.2165
  DDTT: 0.1927
  PSSJ: 0.1815
  WBYY: 0.1812

df_cluster_centres = pd.DataFrame(feature_centres, columns=df_x.columns)
top_features = df_cluster_centres.var().nlargest(5)
for feature, variance in top_features.items():
    print(f"{feature}: {variance:.4e}")
import matplotlib.pyplot as plt
plt.figure(figsize=(12, 6))
top_features.plot(kind='bar')
plt.xlabel('Features')
plt.ylabel('Variance across cluster centers')
plt.xticks(rotation=45, ha='right')
plt.tight_layout()
plt.show()

GXSO: 3.8334e+14
MHEC: 3.8279e+14
PMLC: 3.5513e+14
BDGK: 3.4490e+14
GWEE: 3.1649e+14

kmeans = KMeans(n_clusters=8)
df_top_features = df_no_outliers[top_features.keys()]
cluster_feature_predict = kmeans.fit_predict(df_top_features)
df_feature_predict = df_top_features.copy()
df_feature_predict["cluster"] = cluster_feature_predict.astype(str)
sns.pairplot(df_feature_predict, hue='cluster')

<seaborn.axisgrid.PairGrid at 0x290c65f6ad0>

fig, axes = plt.subplots(4, 2, figsize=(20, 30))
for i, ax in enumerate(axes.flatten()):
    df_cluster_centres.iloc[i].plot(kind='bar', ax=ax)
    ax.set_title(f'Cluster {i}')
    ax.set_xlabel('Features')
    ax.set_ylabel('Value')
    ax.tick_params(axis='x', rotation=90)

plt.tight_layout()
plt.show()

from sklearn.model_selection import train_test_split
X = df_no_outliers.iloc[:, :40]
Y = df_no_outliers.iloc[:, 40:]
print(X.shape, Y.shape)
X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size=0.2)
print(X_train.shape, X_test.shape, Y_train.shape, Y_test.shape)

(4976, 40) (4976, 5)
(3980, 40) (996, 40) (3980, 5) (996, 5)

from sklearn.multioutput import MultiOutputRegressor
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_squared_error

linear = MultiOutputRegressor(LinearRegression())
linear.fit(X_train, Y_train)
Y_train_predict = linear.predict(X_train)
Y_test_predict = linear.predict(X_test)

rmse_train = np.sqrt(mean_squared_error(Y_train, Y_train_predict))
rmse_test = np.sqrt(mean_squared_error(Y_test, Y_test_predict))

print(f"RMSE: Train = {rmse_train:.2f}, Test = {rmse_test:.2f}")

RMSE: Train = 856378361.37, Test = 821665532.05

from sklearn.metrics import r2_score

r2_train = r2_score(Y_train, Y_train_predict)
r2_test = r2_score(Y_test, Y_test_predict)

print(f"R^2: Train = {r2_train:.2f}, Test = {r2_test:.2f}")

R^2: Train = 0.95, Test = 0.96

def adjusted_r2(y_true, y_pred, X):
    r2 = r2_score(y_true, y_pred)
    n = len(y_true)
    p = X.shape[1]
    adjusted_r2 = 1 - (1 - r2) * (n - 1) / (n - p - 1)
    return adjusted_r2

ar2_train = adjusted_r2(Y_train, Y_train_predict, X_train)
ar2_test = adjusted_r2(Y_test, Y_test_predict, X_test)

print(f"Adjusted R^2: Train = {ar2_train:.2f}, Test = {ar2_test:.2f}")

Adjusted R^2: Train = 0.95, Test = 0.95

Y_train_residuals = Y_train - Y_train_predict
fig, axes = plt.subplots(2, 3, figsize=(20, 12))
axes = axes.flatten()

for i, col in enumerate(Y_train.columns):
    sns.scatterplot(x=Y_train[col], y=Y_train_residuals[col], ax=axes[i])
    axes[i].set_title(f'Residuals for {col}')
    axes[i].set_xlabel('Actual Values')
    axes[i].set_ylabel('Residuals')
    axes[i].axhline(y=0, color='r', linestyle='--')
fig.delaxes(axes[5])
plt.tight_layout()
plt.show()

Y_test_residuals = Y_test - Y_test_predict
fig, axes = plt.subplots(2, 3, figsize=(20, 12))
axes = axes.flatten()
for i, col in enumerate(Y_test.columns):
    sns.scatterplot(x=Y_test[col], y=Y_test_residuals[col], ax=axes[i])
    axes[i].set_title(f'Residuals for {col}')
    axes[i].set_xlabel('Actual Values')
    axes[i].set_ylabel('Residuals')
    axes[i].axhline(y=0, color='r', linestyle='--')
fig.delaxes(axes[5])
plt.tight_layout()
plt.show()

from sklearn.preprocessing import PolynomialFeatures
from sklearn.pipeline import make_pipeline

poly = make_pipeline(PolynomialFeatures(degree=2), MultiOutputRegressor(LinearRegression()))
poly.fit(X_train, Y_train)

Y_train_predict_poly = poly.predict(X_train)
Y_test_predict_poly = poly.predict(X_test)

r2_train_poly = r2_score(Y_train, Y_train_predict_poly)
r2_test_poly = r2_score(Y_test, Y_test_predict_poly)

print(f"Polynomial R^2: Train = {r2_train_poly:.2f}, Test = {r2_test_poly:.2f}")

Polynomial R^2: Train = 0.97, Test = 0.94

plt.figure(figsize=(15, 10))
for i, response in enumerate(Y.columns):
    coefficients = linear.estimators_[i].coef_
    plt.subplot(2, 3, i+1)
    plt.bar(X.columns, coefficients)
    plt.title(f'Feature coefficients for {response}')
    plt.xticks(rotation=90)
plt.tight_layout()
plt.show()

for i, response in enumerate(Y.columns):
    coefficients = linear.estimators_[i].coef_
    feature_importance = pd.DataFrame({'feature': X.columns, 'importance': abs(coefficients)})
    feature_importance = feature_importance.sort_values('importance', ascending=False)
    
    print(f"\nTop 5 features for {response}:")
    print(feature_importance.head(5))

Top 5 features for RESP__0:
   feature  importance
17    BDGK      161.15
29    MHEC      131.82
0     ZYTI       97.02
26    VKRW        5.25
13    RNOZ        3.63

Top 5 features for RESP__1:
   feature  importance
7     NANX      218.88
1     WIBD      191.41
19    PFKF      188.92
9     JIEK        6.48
0     ZYTI        5.65

Top 5 features for RESP__2:
   feature  importance
2     KVJI       95.31
25    GQUZ       64.73
12    RLCO       52.87
39    XYKI        2.73
30    TUXT        1.54

Top 5 features for RESP__3:
   feature  importance
5     RXUY      191.41
35    ROQE      161.49
21    XBUG       82.92
20    ZKOY        5.38
0     ZYTI        4.65

Top 5 features for RESP__4:
   feature  importance
16    WBYY      200.83
31    UZSD      176.14
4     OFSQ      117.62
24    YKJO        4.56
37    RTDN        4.50

corr_matrix = X.corr()
plt.figure(figsize=(12, 10))
sns.heatmap(corr_matrix, annot=False, cmap='coolwarm')
plt.show()

from statsmodels.stats.outliers_influence import variance_inflation_factor

df_vif = pd.DataFrame()
df_vif["feature"] = X.columns
df_vif["VIF"] = [variance_inflation_factor(X.values, i) for i in range(X.shape[1])]
print(df_vif)

   feature     VIF
0     ZYTI   71.65
1     WIBD   58.06
2     KVJI  114.83
3     CSTT   74.98
4     OFSQ   76.43
5     RXUY   71.27
6     GWEE   73.37
7     NANX   82.44
8     PMLC   74.90
9     JIEK   88.54
10    GXSO   81.20
11    PTAX   92.10
12    RLCO   70.60
13    RNOZ   68.59
14    QRWO   56.84
15    PSSJ   80.89
16    WBYY   60.84
17    BDGK   90.20
18    FWKW   78.42
19    PFKF   68.51
20    ZKOY  109.87
21    XBUG   67.33
22    OMXX   95.29
23    IXKB   93.14
24    YKJO   76.65
25    GQUZ   65.33
26    VKRW   79.50
27    DDTT   81.84
28    QXYM   87.60
29    MHEC   60.19
30    TUXT   79.26
31    UZSD   58.91
32    AOQU   72.78
33    RJWA   73.98
34    VTQI   70.22
35    ROQE   67.32
36    ZMHO   70.31
37    RTDN   68.10
38    MJXP   91.14
39    XYKI   91.28

from sklearn.linear_model import LassoCV

lasso_cv = MultiOutputRegressor(LassoCV(cv=5))
lasso_cv.fit(X_train, Y_train)

Y_train_predict = lasso_cv.predict(X_train)
Y_test_predict = lasso_cv.predict(X_test)

rmse_train = np.sqrt(mean_squared_error(Y_train, Y_train_predict))
rmse_test = np.sqrt(mean_squared_error(Y_test, Y_test_predict))

print(f"RMSE: Train = {rmse_train:.2f}, Test = {rmse_test:.2f}")

ar2_train = adjusted_r2(Y_train, Y_train_predict, X_train)
ar2_test = adjusted_r2(Y_test, Y_test_predict, X_test)

print(f"Adjusted R^2: Train = {ar2_train:.2f}, Test = {ar2_test:.2f}")

for i, estimator in enumerate(lasso_cv.estimators_):
    print(f"Optimal lambda for RESP__{i}: {estimator.alpha_:.4f}")

for i, estimator in enumerate(lasso_cv.estimators_):
    print(f"\nNon-zero coefficients for RESP__{i}:")
    non_zero = [(feature, coef) for feature, coef in zip(X_train.columns, estimator.coef_) if coef != 0]
    for feature, coef in non_zero:
        print(f"{feature}: {coef:.4f}")

RMSE: Train = 858979154.78, Test = 819671227.40
Adjusted R^2: Train = 0.95, Test = 0.95
Optimal lambda for RESP__0: 211072329325221.4375
Optimal lambda for RESP__1: 143528779147763.8125
Optimal lambda for RESP__2: 115274539055589.2656
Optimal lambda for RESP__3: 231733766359050.7500
Optimal lambda for RESP__4: 281556664206038.6250

Non-zero coefficients for RESP__0:
ZYTI: 97.5460
GXSO: -1.4158
RNOZ: -2.5652
WBYY: -1.1718
BDGK: 162.9124
FWKW: 0.3984
PFKF: -0.1909
IXKB: -0.8350
GQUZ: -0.0359
VKRW: 3.3633
QXYM: 1.0138
MHEC: 131.8632
XYKI: -1.4913

Non-zero coefficients for RESP__1:
ZYTI: -3.9190
WIBD: 189.1183
RXUY: -1.7726
GWEE: 2.1716
NANX: 215.7693
JIEK: -2.9410
GXSO: -3.5772
RLCO: -0.7370
RNOZ: 0.2055
BDGK: -0.4563
PFKF: 189.3562
XBUG: 1.4155
IXKB: -0.5115
VKRW: 0.0171
QXYM: -0.2934
MHEC: -0.5729
UZSD: -0.5481
VTQI: 0.1789
ROQE: -0.3620

Non-zero coefficients for RESP__2:
KVJI: 95.2080
NANX: 1.4413
PTAX: 0.3451
RLCO: 51.7411
RNOZ: 0.1362
WBYY: 0.1605
BDGK: -1.6324
FWKW: -0.1884
XBUG: -1.1690
IXKB: 0.4914
YKJO: -0.0921
GQUZ: 64.6574
DDTT: -0.7156
TUXT: 1.1777
XYKI: 1.4485

Non-zero coefficients for RESP__3:
ZYTI: 1.6113
RXUY: 192.6434
NANX: -1.2326
PMLC: 2.2817
JIEK: -3.3281
QRWO: 2.4221
PSSJ: 1.8637
FWKW: 3.3602
ZKOY: 3.8594
XBUG: 81.9394
OMXX: 0.0284
GQUZ: -1.0103
TUXT: -0.0756
ROQE: 161.2673
ZMHO: -0.8405

Non-zero coefficients for RESP__4:
KVJI: 0.7826
OFSQ: 114.8221
NANX: 0.6402
PMLC: -2.6259
WBYY: 202.2977
YKJO: -4.5653
VKRW: 0.4209
MHEC: -1.3507
UZSD: 176.9233
RTDN: -2.6667

plt.figure(figsize=(15, 10))
for i, response in enumerate(Y.columns):
    coefficients = lasso_cv.estimators_[i].coef_
    plt.subplot(2, 3, i+1)
    plt.bar(X.columns, coefficients)
    plt.title(f'Feature coefficients for {response}')
    plt.xticks(rotation=90)
plt.tight_layout()
plt.show()

for i, response in enumerate(Y.columns):
    coefficients = lasso_cv.estimators_[i].coef_
    feature_importance = pd.DataFrame({'feature': X.columns, 'importance': abs(coefficients)})
    feature_importance = feature_importance.sort_values('importance', ascending=False)
    
    print(f"\nTop 5 features for {response}:")
    print(feature_importance.head(5))

Top 5 features for RESP__0:
   feature  importance
17    BDGK      162.91
29    MHEC      131.86
0     ZYTI       97.55
26    VKRW        3.36
13    RNOZ        2.57

Top 5 features for RESP__1:
   feature  importance
7     NANX      215.77
19    PFKF      189.36
1     WIBD      189.12
0     ZYTI        3.92
10    GXSO        3.58

Top 5 features for RESP__2:
   feature  importance
2     KVJI       95.21
25    GQUZ       64.66
12    RLCO       51.74
17    BDGK        1.63
39    XYKI        1.45

Top 5 features for RESP__3:
   feature  importance
5     RXUY      192.64
35    ROQE      161.27
21    XBUG       81.94
20    ZKOY        3.86
18    FWKW        3.36

Top 5 features for RESP__4:
   feature  importance
16    WBYY      202.30
31    UZSD      176.92
4     OFSQ      114.82
24    YKJO        4.57
37    RTDN        2.67

pca = PCA(n_components=7)
np_pca = pca.fit_transform(df_x)
df_pca = pd.DataFrame(np_pca, columns=[f"PC{i}" for i in range(1, 8)])

kmeans = KMeans(n_clusters=8)
cluster_predict = kmeans.fit_predict(np_pca)

df_clustered = df_no_outliers.copy()
df_clustered['cluster'] = cluster_predict.astype(str)
cluster_means = df_clustered.groupby('cluster')[Y.columns].mean()

plt.figure(figsize=(15, 10))
for i, response in enumerate(Y.columns):
    plt.subplot(2, 3, i+1)
    cluster_means[response].plot(kind='bar')
    plt.title(f'Mean {response} by Cluster')
    plt.xlabel('Cluster')
    plt.ylabel('Mean Response')
plt.tight_layout()
plt.show()

cluster_means_all = df_clustered.groupby('cluster').mean()

cluster_means_all['cluster'] = cluster_means_all.index

X_cluster = cluster_means_all.iloc[:, :40]  
Y_cluster = cluster_means_all.iloc[:, 40:45] 

print(X_cluster.shape, Y_cluster.shape)

(8, 40) (8, 5)

model = LinearRegression()
model.fit(X_cluster, Y_cluster)

Y_pred = model.predict(X_cluster)

r2_scores = r2_score(Y_cluster, Y_pred, multioutput='raw_values')

for i, response in enumerate(Y_cluster.columns):
    print(f"R-squared for {response}: {r2_scores[i]:.4f}")

R-squared for RESP__0: 1.0000
R-squared for RESP__1: 1.0000
R-squared for RESP__2: 1.0000
R-squared for RESP__3: 1.0000
R-squared for RESP__4: 1.0000

fig, axes = plt.subplots(2, 3, figsize=(20, 12))
axes = axes.flatten()

for i, response in enumerate(Y_cluster.columns):
    axes[i].scatter(Y_pred[:, i], Y_cluster[response])
    axes[i].plot([Y_pred[:, i].min(), Y_pred[:, i].max()], 
                 [Y_pred[:, i].min(), Y_pred[:, i].max()], 
                 'r--', lw=2)
    axes[i].set_xlabel(f'Predicted {response}')
    axes[i].set_ylabel(f'Actual {response}')
    axes[i].set_title(f'{response}: R² = {r2_scores[i]:.4f}')

fig.delaxes(axes[5])
plt.tight_layout()
plt.show()

	ZYTI	WIBD	KVJI	CSTT	OFSQ	RXUY	GWEE	NANX	PMLC	JIEK	GXSO	PTAX	RLCO	RNOZ	QRWO	PSSJ	WBYY	BDGK	FWKW	PFKF	ZKOY	XBUG	OMXX	IXKB	YKJO	GQUZ	VKRW	DDTT	QXYM	MHEC	TUXT	UZSD	AOQU	RJWA	VTQI	ROQE	ZMHO	RTDN	MJXP	XYKI	RESP__0	RESP__1	RESP__2	RESP__3	RESP__4
0	7.12e+07	66780003.23249369	98615834.8241663	52159482.56117151	82784798.00617953	6.06e+07	7.50e+07	8.37e+07	5.02e+07	67148304.13857159	7.71e+07	8.27e+07	68347063.23106696	86863631.66238013	5.14e+07	76935314.59720963	105443191.56660278	8.91e+07	7.79e+07	6.44e+07	8.09e+07	6.33e+07	9.81e+07	8.40e+07	5.31e+07	6.47e+07	1.03e+08	7.59e+07	8.31e+07	1.02e+08	73393667.46745682	8.61e+07	6.40e+07	8.56e+07	7.01e+07	93842606.08280182	65363263.588700525	54351366.2627504	86923737.15743968	7.79e+07	36519700861.36076	46449066259.069855	1.87e+10	3.25e+10	4.68e+10
1	5.80e+07	52206572.93108849	98309387.02801965	64901682.897473596	57695764.36746716	6.96e+07	9.30e+07	9.51e+07	6.38e+07	91303795.85450855	1.12e+08	9.82e+07	64604823.260036215	81374620.04454753	5.88e+07	89711575.45912442	39303004.81084073	5.97e+07	5.60e+07	-1.33e+11	8.82e+07	6.01e+07	6.57e+07	1.06e+08	8.70e+07	9.74e+07	8.68e+07	7.24e+07	8.95e+07	8.75e+07	95533827.80156274	4.20e+07	5.67e+07	5.69e+07	4.81e+07	58506159.2007904	97113018.54325898	94065372.48496212	91826626.8781961	7.27e+07	28442049767.828587	42252211000.022964	2.06e+10	2.85e+10	2.36e+10
2	8.93e+07	59058481.656934984	90340950.98559198	68251937.91355611	89272597.48780008	6.70e+07	8.26e+07	9.48e+07	6.09e+07	53998583.56195885	6.46e+07	8.04e+07	59588103.00683621	86360346.40430261	7.96e+07	86348427.75161901	99173173.9837645	9.37e+07	7.78e+07	5.12e+07	8.20e+07	5.92e+07	9.50e+07	7.23e+07	4.03e+07	8.50e+07	9.20e+07	6.11e+07	8.26e+07	7.44e+07	71433974.72584881	8.37e+07	6.96e+07	7.38e+07	6.12e+07	91553527.28947656	66397909.15593497	42828943.43174528	86023891.74409276	8.09e+07	33984056943.228825	43317034961.763405	1.89e+10	3.40e+10	4.50e+10
3	4.16e+07	87279722.31068213	101620049.21358526	90873708.86792825	94655232.50324139	4.40e+07	7.62e+07	6.79e+07	4.83e+07	68304099.42017053	8.87e+07	1.00e+08	70052071.93627621	68627709.75841182	7.84e+07	57881674.17562354	71188279.49110305	1.01e+08	5.18e+07	6.28e+07	6.96e+07	6.93e+07	5.59e+07	7.63e+07	8.12e+07	4.76e+07	6.99e+07	9.79e+07	7.22e+07	8.70e+07	58931006.120989144	5.62e+07	8.80e+07	9.06e+07	5.54e+07	64591587.78148231	90276647.66245224	56330658.069345966	81016608.49633285	7.24e+07	33374977724.665	45330357093.554	1.76e+10	2.58e+10	3.65e+10
4	9.06e+07	57388025.488219306	57568362.19853688	75526475.58607598	60713729.34205614	8.32e+07	8.56e+07	8.17e+07	9.73e+07	78009778.11793171	5.50e+07	5.34e+07	50981332.5625033	65076827.86445975	5.69e+07	97149817.50893311	58031957.693665884	9.69e+07	1.02e+08	7.51e+07	9.08e+07	8.59e+07	8.57e+07	7.25e+07	8.03e+07	5.96e+07	7.35e+07	8.16e+07	8.78e+07	8.58e+07	45921221.88581882	6.90e+07	5.86e+07	5.40e+07	7.97e+07	72331407.63715056	60207543.17420656	57404939.27796871	91207189.35157438	6.98e+07	36076115306.28484	45277780420.7335	1.25e+10	3.51e+10	3.20e+10
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
4995	7.41e+07	58262507.06873764	92701179.7179384	67796505.77780084	60996313.33816352	1.05e+08	3.67e+07	6.85e+07	9.24e+07	53654575.16920183	1.07e+08	9.00e+07	85839241.93596187	77902522.01750344	5.84e+07	86443114.50503628	57347972.89513925	6.48e+07	7.46e+07	7.01e+07	9.28e+07	6.57e+07	8.50e+07	8.40e+07	7.50e+07	6.17e+07	6.62e+07	6.38e+07	8.63e+07	5.53e+07	70640478.52763283	6.71e+07	5.83e+07	6.62e+07	7.87e+07	80479212.19958583	44443060.7246354	89698586.36942211	49768664.34920095	9.32e+07	26309568347.696503	40888684838.742096	1.80e+10	4.00e+10	3.16e+10
4996	6.63e+07	44913372.921941906	93559494.10271989	54065530.2835989	56083538.15939477	6.06e+07	7.73e+07	9.62e+07	6.80e+07	99407331.31693117	9.73e+07	9.74e+07	61454056.53538433	89855424.75312191	5.03e+07	71468658.62508146	40146847.83650946	4.38e+07	5.80e+07	6.52e+07	8.45e+07	6.16e+07	8.41e+07	9.70e+07	9.17e+07	8.78e+07	7.29e+07	6.11e+07	7.86e+07	9.54e+07	87501383.37127593	5.88e+07	6.10e+07	5.85e+07	6.23e+07	53894275.853690684	84042368.72940478	99886500.6802299	86733756.31355335	7.94e+07	26512646628.503376	45338586455.458115	1.77e+10	2.63e+10	2.55e+10
4997	9.74e+07	51594995.715354934	65552552.49277553	98008359.76560771	76960297.88906516	8.18e+07	6.96e+07	8.30e+07	8.47e+07	90739941.20183507	4.73e+07	7.65e+07	93275227.52404302	52439119.3751166	5.20e+07	72569921.86725812	71439253.9880689	8.46e+07	7.55e+07	6.68e+07	8.73e+07	6.73e+07	5.60e+07	7.02e+07	5.67e+07	7.00e+07	6.66e+07	7.50e+07	9.30e+07	5.88e+07	108437382.89313962	8.32e+07	9.03e+07	5.24e+07	4.76e+07	56896935.732826635	75933653.90980247	50265174.73736608	37565629.21780636	6.63e+07	31568811420.915096	42070391054.60378	1.67e+10	3.24e+10	4.00e+10
4998	5.86e+07	66843238.02653567	71275786.0738799	55771212.44004866	72847960.13200027	1.11e+08	5.37e+07	6.26e+07	9.16e+07	58169036.77614904	9.01e+07	8.24e+07	84858524.17996946	78123267.90743856	6.47e+07	95093568.50489672	64316317.91820028	5.93e+07	7.50e+07	6.05e+07	1.04e+08	7.75e+07	8.46e+07	8.43e+07	7.87e+07	4.88e+07	6.26e+07	5.01e+07	6.34e+07	5.03e+07	65383997.971619956	5.64e+07	6.89e+07	5.36e+07	6.96e+07	85813857.88571435	47238535.41352343	99096972.99522956	55161687.71002631	7.62e+07	22176601413.23164	39278765535.841156	1.53e+10	4.17e+10	3.32e+10
4999	5.58e+07	76902968.80040008	82714128.91204472	55603146.54898293	75568004.44787523	6.76e+07	7.99e+07	9.47e+07	4.97e+07	61735599.656930335	6.81e+07	9.00e+07	70151739.14748453	82451061.22957435	5.02e+07	79849349.2104344	95368848.53425784	9.76e+07	8.90e+07	5.13e+07	8.06e+07	6.18e+07	9.08e+07	8.28e+07	4.36e+07	7.56e+07	9.90e+07	6.68e+07	8.97e+07	8.06e+07	79624087.37402406	1.03e+08	6.74e+07	7.41e+07	6.71e+07	98605618.73921068	56446409.76727988	54678415.73510831	99165159.52924104	7.37e+07	32710388548.81561	45508803880.13961	1.75e+10	3.57e+10	4.69e+10

	ZYTI	WIBD	KVJI	CSTT	OFSQ	RXUY	GWEE	NANX	PMLC	JIEK	GXSO	PTAX	RLCO	RNOZ	QRWO	PSSJ	WBYY	BDGK	FWKW	PFKF	ZKOY	XBUG	OMXX	IXKB	YKJO	GQUZ	VKRW	DDTT	QXYM	MHEC	TUXT	UZSD	AOQU	RJWA	VTQI	ROQE	ZMHO	RTDN	MJXP	XYKI	RESP__0	RESP__1	RESP__2	RESP__3	RESP__4
count	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03	4.98e+03
mean	1.10e+08	9.11e+07	3.48e+07	-3.23e+07	7.43e+07	1.14e+08	4.79e+07	7.52e+07	9.59e+07	8.62e+07	5.22e+07	3.91e+07	1.02e+08	6.73e+07	6.10e+07	3.04e+07	9.50e+07	4.37e+07	1.42e+08	4.42e+07	1.11e+08	9.23e+07	2.03e+08	7.04e+07	7.64e+07	1.06e+08	5.04e+07	7.97e+07	5.29e+07	7.35e+07	6.42e+07	6.55e+07	4.23e+07	8.96e+07	8.30e+06	7.12e+07	6.90e+07	3.94e+07	4.59e+07	1.87e+06	3.08e+10	4.57e+10	1.70e+10	3.25e+10	3.49e+10
std	3.71e+09	3.31e+09	2.40e+09	4.24e+09	2.61e+09	4.25e+09	1.91e+09	3.63e+09	1.56e+09	3.31e+09	3.51e+09	3.87e+09	2.29e+09	2.72e+09	1.18e+07	3.78e+09	4.55e+09	2.68e+09	3.57e+09	1.88e+09	2.13e+09	4.82e+09	4.53e+09	3.27e+09	2.67e+09	2.55e+09	1.62e+09	2.81e+09	3.92e+09	3.82e+09	5.56e+09	1.64e+07	2.79e+09	1.69e+09	4.98e+09	3.94e+09	2.19e+09	2.05e+09	2.31e+09	4.92e+09	4.62e+09	4.83e+09	4.47e+09	5.45e+09	6.03e+09
min	-1.34e+11	-1.44e+11	-1.21e+11	-1.92e+11	-1.15e+11	-1.54e+11	-1.35e+11	-1.86e+11	3.15e+07	-1.50e+11	-1.50e+11	-1.82e+11	3.02e+07	-1.45e+11	0.00e+00	-1.97e+11	-1.33e+11	-1.89e+11	3.20e+07	-1.33e+11	5.13e+07	-1.58e+11	3.91e+07	-1.84e+11	-1.17e+11	0.00e+00	-1.14e+11	-1.35e+11	-1.60e+11	-1.89e+11	-1.87e+11	0.00e+00	-1.26e+11	3.00e+07	-1.98e+11	-1.94e+11	-1.17e+11	-1.45e+11	-1.63e+11	-1.96e+11	0.00e+00	3.19e+10	-1.72e+11	2.18e+10	2.03e+10
25%	6.25e+07	5.50e+07	7.21e+07	5.88e+07	6.05e+07	5.83e+07	5.86e+07	6.45e+07	5.69e+07	6.83e+07	5.67e+07	6.28e+07	5.63e+07	5.66e+07	5.25e+07	5.78e+07	5.93e+07	6.37e+07	5.90e+07	5.59e+07	7.28e+07	5.85e+07	6.39e+07	6.88e+07	6.07e+07	5.91e+07	6.03e+07	6.21e+07	6.91e+07	5.49e+07	6.36e+07	5.34e+07	5.72e+07	5.49e+07	5.89e+07	5.80e+07	6.14e+07	5.58e+07	6.49e+07	6.82e+07	2.73e+10	4.31e+10	1.59e+10	2.85e+10	3.00e+10
50%	7.17e+07	6.52e+07	8.50e+07	6.73e+07	6.95e+07	6.75e+07	7.67e+07	7.72e+07	7.32e+07	8.32e+07	7.04e+07	8.11e+07	6.51e+07	7.18e+07	6.02e+07	7.13e+07	6.79e+07	8.76e+07	6.74e+07	6.72e+07	8.00e+07	6.63e+07	7.68e+07	8.06e+07	7.06e+07	7.03e+07	7.18e+07	7.85e+07	7.88e+07	7.26e+07	7.43e+07	6.16e+07	6.71e+07	6.31e+07	6.74e+07	6.94e+07	7.11e+07	6.27e+07	8.15e+07	7.67e+07	3.12e+10	4.55e+10	1.72e+10	3.09e+10	3.50e+10
75%	8.29e+07	8.13e+07	9.55e+07	8.57e+07	7.76e+07	8.65e+07	8.97e+07	8.91e+07	9.09e+07	9.26e+07	9.38e+07	9.29e+07	8.18e+07	8.55e+07	6.94e+07	8.51e+07	8.09e+07	9.70e+07	8.05e+07	8.74e+07	8.94e+07	7.57e+07	8.64e+07	9.01e+07	7.91e+07	8.13e+07	8.68e+07	9.25e+07	8.60e+07	9.12e+07	9.15e+07	7.65e+07	8.23e+07	7.35e+07	7.72e+07	8.27e+07	8.34e+07	8.15e+07	9.25e+07	8.42e+07	3.42e+10	4.80e+10	1.84e+10	3.65e+10	3.86e+10
max	1.79e+11	1.48e+11	1.22e+08	1.19e+08	1.43e+11	1.89e+11	1.20e+08	1.76e+11	1.10e+11	1.79e+11	1.58e+11	1.36e+11	1.62e+11	1.26e+11	1.06e+08	1.23e+11	1.91e+11	1.25e+08	1.79e+11	1.25e+08	1.50e+11	1.76e+11	1.91e+11	1.39e+11	1.47e+11	1.80e+11	1.19e+08	1.45e+11	1.79e+11	1.92e+11	1.78e+11	1.20e+08	1.02e+11	1.20e+11	1.28e+11	1.98e+11	1.01e+11	1.18e+08	1.21e+08	1.93e+11	4.40e+10	1.92e+11	1.27e+11	1.08e+11	1.51e+11

	ZYTI	WIBD	KVJI	CSTT	OFSQ	RXUY	GWEE	NANX	PMLC	JIEK	GXSO	PTAX	RLCO	RNOZ	QRWO	PSSJ	WBYY	BDGK	FWKW	PFKF	ZKOY	XBUG	OMXX	IXKB	YKJO	GQUZ	VKRW	DDTT	QXYM	MHEC	TUXT	UZSD	AOQU	RJWA	VTQI	ROQE	ZMHO	RTDN	MJXP	XYKI
0	7.12e+07	6.68e+07	9.86e+07	5.22e+07	8.28e+07	6.06e+07	7.50e+07	8.37e+07	5.02e+07	6.71e+07	7.71e+07	8.27e+07	6.83e+07	8.69e+07	5.14e+07	7.69e+07	1.05e+08	8.91e+07	7.79e+07	6.44e+07	8.09e+07	6.33e+07	9.81e+07	8.40e+07	5.31e+07	6.47e+07	1.03e+08	7.59e+07	8.31e+07	1.02e+08	7.34e+07	8.61e+07	6.40e+07	8.56e+07	7.01e+07	9.38e+07	6.54e+07	5.44e+07	8.69e+07	7.79e+07
1	5.80e+07	5.22e+07	9.83e+07	6.49e+07	5.77e+07	6.96e+07	9.30e+07	9.51e+07	6.38e+07	9.13e+07	1.12e+08	9.82e+07	6.46e+07	8.14e+07	5.88e+07	8.97e+07	3.93e+07	5.97e+07	5.60e+07	6.72e+07	8.82e+07	6.01e+07	6.57e+07	1.06e+08	8.70e+07	9.74e+07	8.68e+07	7.24e+07	8.95e+07	8.75e+07	9.55e+07	4.20e+07	5.67e+07	5.69e+07	4.81e+07	5.85e+07	9.71e+07	9.41e+07	9.18e+07	7.27e+07
2	8.93e+07	5.91e+07	9.03e+07	6.83e+07	8.93e+07	6.70e+07	8.26e+07	9.48e+07	6.09e+07	5.40e+07	6.46e+07	8.04e+07	5.96e+07	8.64e+07	7.96e+07	8.63e+07	9.92e+07	9.37e+07	7.78e+07	5.12e+07	8.20e+07	5.92e+07	9.50e+07	7.23e+07	4.03e+07	8.50e+07	9.20e+07	6.11e+07	8.26e+07	7.44e+07	7.14e+07	8.37e+07	6.96e+07	7.38e+07	6.12e+07	9.16e+07	6.64e+07	4.28e+07	8.60e+07	8.09e+07
3	4.16e+07	8.73e+07	1.02e+08	9.09e+07	9.47e+07	4.40e+07	7.62e+07	6.79e+07	4.83e+07	6.83e+07	8.87e+07	1.00e+08	7.01e+07	6.86e+07	7.84e+07	5.79e+07	7.12e+07	1.01e+08	5.18e+07	6.28e+07	6.96e+07	6.93e+07	5.59e+07	7.63e+07	8.12e+07	4.76e+07	6.99e+07	9.79e+07	7.22e+07	8.70e+07	5.89e+07	5.62e+07	8.80e+07	9.06e+07	5.54e+07	6.46e+07	9.03e+07	5.63e+07	8.10e+07	7.24e+07
4	9.06e+07	5.74e+07	5.76e+07	7.55e+07	6.07e+07	8.32e+07	8.56e+07	8.17e+07	9.73e+07	7.80e+07	5.50e+07	5.34e+07	5.10e+07	6.51e+07	5.69e+07	9.71e+07	5.80e+07	9.69e+07	1.02e+08	7.51e+07	9.08e+07	8.59e+07	8.57e+07	7.25e+07	8.03e+07	5.96e+07	7.35e+07	8.16e+07	8.78e+07	8.58e+07	4.59e+07	6.90e+07	5.86e+07	5.40e+07	7.97e+07	7.23e+07	6.02e+07	5.74e+07	9.12e+07	6.98e+07
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
4995	7.41e+07	5.83e+07	9.27e+07	6.78e+07	6.10e+07	1.05e+08	3.67e+07	6.85e+07	9.24e+07	5.37e+07	1.07e+08	9.00e+07	8.58e+07	7.79e+07	5.84e+07	8.64e+07	5.73e+07	6.48e+07	7.46e+07	7.01e+07	9.28e+07	6.57e+07	8.50e+07	8.40e+07	7.50e+07	6.17e+07	6.62e+07	6.38e+07	8.63e+07	5.53e+07	7.06e+07	6.71e+07	5.83e+07	6.62e+07	7.87e+07	8.05e+07	4.44e+07	8.97e+07	4.98e+07	9.32e+07
4996	6.63e+07	4.49e+07	9.36e+07	5.41e+07	5.61e+07	6.06e+07	7.73e+07	9.62e+07	6.80e+07	9.94e+07	9.73e+07	9.74e+07	6.15e+07	8.99e+07	5.03e+07	7.15e+07	4.01e+07	4.38e+07	5.80e+07	6.52e+07	8.45e+07	6.16e+07	8.41e+07	9.70e+07	9.17e+07	8.78e+07	7.29e+07	6.11e+07	7.86e+07	9.54e+07	8.75e+07	5.88e+07	6.10e+07	5.85e+07	6.23e+07	5.39e+07	8.40e+07	9.99e+07	8.67e+07	7.94e+07
4997	9.74e+07	5.16e+07	6.56e+07	9.80e+07	7.70e+07	8.18e+07	6.96e+07	8.30e+07	8.47e+07	9.07e+07	4.73e+07	7.65e+07	9.33e+07	5.24e+07	5.20e+07	7.26e+07	7.14e+07	8.46e+07	7.55e+07	6.68e+07	8.73e+07	6.73e+07	5.60e+07	7.02e+07	5.67e+07	7.00e+07	6.66e+07	7.50e+07	9.30e+07	5.88e+07	1.08e+08	8.32e+07	9.03e+07	5.24e+07	4.76e+07	5.69e+07	7.59e+07	5.03e+07	3.76e+07	6.63e+07
4998	5.86e+07	6.68e+07	7.13e+07	5.58e+07	7.28e+07	1.11e+08	5.37e+07	6.26e+07	9.16e+07	5.82e+07	9.01e+07	8.24e+07	8.49e+07	7.81e+07	6.47e+07	9.51e+07	6.43e+07	5.93e+07	7.50e+07	6.05e+07	1.04e+08	7.75e+07	8.46e+07	8.43e+07	7.87e+07	4.88e+07	6.26e+07	5.01e+07	6.34e+07	5.03e+07	6.54e+07	5.64e+07	6.89e+07	5.36e+07	6.96e+07	8.58e+07	4.72e+07	9.91e+07	5.52e+07	7.62e+07
4999	5.58e+07	7.69e+07	8.27e+07	5.56e+07	7.56e+07	6.76e+07	7.99e+07	9.47e+07	4.97e+07	6.17e+07	6.81e+07	9.00e+07	7.02e+07	8.25e+07	5.02e+07	7.98e+07	9.54e+07	9.76e+07	8.90e+07	5.13e+07	8.06e+07	6.18e+07	9.08e+07	8.28e+07	4.36e+07	7.56e+07	9.90e+07	6.68e+07	8.97e+07	8.06e+07	7.96e+07	1.03e+08	6.74e+07	7.41e+07	6.71e+07	9.86e+07	5.64e+07	5.47e+07	9.92e+07	7.37e+07

	PC1	PC2	PC3	PC4	PC5	PC6	PC7
ZYTI	-1.74e-01	0.12	1.39e-01	-4.81e-02	2.44e-01	6.62e-02	-1.06e-01
WIBD	5.63e-02	-0.19	-1.19e-01	3.15e-01	-6.31e-02	-4.31e-02	-2.69e-01
KVJI	2.96e-01	-0.09	-5.10e-02	1.53e-01	9.53e-03	-5.75e-02	1.22e-01
CSTT	-7.34e-02	-0.06	-3.27e-01	-6.92e-02	1.45e-01	1.93e-01	2.63e-02
OFSQ	3.32e-02	-0.17	-1.57e-01	6.76e-02	3.65e-02	-9.38e-02	7.54e-02
RXUY	-1.60e-01	0.35	-3.48e-02	6.99e-02	-8.01e-03	2.84e-02	-6.39e-02
GWEE	9.70e-02	-0.09	3.72e-01	-1.08e-01	1.67e-01	-2.01e-02	-2.53e-01
NANX	1.03e-01	0.09	1.44e-01	-1.91e-01	1.22e-01	1.34e-01	2.70e-01
PMLC	-2.31e-01	0.33	-1.35e-01	-9.20e-02	5.10e-02	6.82e-02	8.98e-04
JIEK	1.38e-02	-0.17	-6.34e-03	-3.12e-01	2.28e-03	-4.00e-02	-2.35e-01
GXSO	3.98e-01	0.09	-1.34e-01	1.35e-01	-9.57e-02	2.12e-03	-8.10e-02
PTAX	3.21e-01	0.05	-1.32e-01	8.92e-02	1.46e-01	1.28e-01	-7.26e-02
RLCO	4.77e-03	0.27	-1.29e-01	5.29e-02	3.13e-01	-2.66e-02	1.61e-02
RNOZ	2.24e-01	0.05	2.33e-01	2.77e-02	1.04e-01	-3.23e-01	3.16e-02
QRWO	3.25e-02	-0.01	-1.27e-01	1.92e-01	3.59e-02	-1.00e-01	-1.11e-01
PSSJ	6.61e-03	0.27	1.76e-01	1.76e-01	-7.91e-02	5.62e-02	-1.05e-01
WBYY	-6.07e-02	-0.16	2.34e-02	3.98e-02	3.56e-02	-4.89e-02	4.84e-01
BDGK	-2.50e-01	-0.24	8.38e-02	1.59e-01	1.70e-01	2.15e-01	-1.14e-01
FWKW	-1.36e-01	0.09	2.16e-01	2.50e-01	-6.66e-02	1.30e-01	-2.61e-04
PFKF	-2.86e-01	-0.04	1.67e-02	-8.24e-02	-2.34e-01	-3.03e-01	2.29e-02
ZKOY	1.38e-02	0.19	5.45e-02	3.94e-02	-4.88e-02	-5.05e-02	-6.12e-02
XBUG	-1.79e-01	0.08	7.75e-02	-9.10e-03	-7.14e-02	-5.47e-02	-2.16e-01
OMXX	-1.09e-01	0.03	1.90e-01	1.89e-01	9.65e-02	-1.80e-01	-1.14e-01
IXKB	2.32e-01	0.07	2.96e-03	-1.99e-02	-1.06e-01	1.08e-01	2.09e-01
YKJO	8.94e-02	0.09	1.60e-02	-1.38e-01	-2.08e-01	-3.66e-02	-2.50e-01
GQUZ	5.26e-02	-0.06	1.80e-01	-2.16e-01	-3.59e-02	-1.42e-01	2.11e-01
VKRW	5.48e-02	-0.19	3.44e-01	9.22e-02	6.27e-02	-2.90e-02	-9.54e-02
DDTT	-1.64e-01	-0.24	-1.87e-01	-1.37e-01	-1.85e-01	5.88e-02	-1.26e-01
QXYM	5.96e-02	0.18	3.95e-02	1.22e-02	6.37e-02	2.71e-01	-3.94e-02
MHEC	1.64e-01	-0.03	1.39e-01	-4.26e-02	-2.68e-01	5.24e-01	-1.25e-01
TUXT	1.41e-01	-0.07	7.33e-02	-1.68e-01	3.60e-01	-1.21e-01	-1.80e-01
UZSD	-2.94e-02	-0.11	1.15e-01	9.94e-02	3.40e-01	1.85e-01	1.94e-01
AOQU	5.67e-03	-0.20	-1.61e-01	1.45e-01	1.88e-01	-5.05e-02	-2.01e-01
RJWA	7.80e-02	-0.21	-1.02e-01	2.08e-01	-1.10e-01	6.99e-02	-3.01e-02
VTQI	-8.94e-02	0.07	-5.29e-02	1.10e-01	-2.36e-01	-6.75e-02	9.05e-02
ROQE	2.74e-02	0.05	1.05e-01	3.83e-01	-7.80e-02	2.74e-02	1.28e-01
ZMHO	1.31e-01	-0.08	-1.51e-02	-3.32e-01	-6.93e-02	1.37e-01	-1.60e-02
RTDN	2.62e-01	0.23	-3.68e-02	-8.40e-02	-8.62e-02	-1.97e-01	-1.25e-01
MJXP	-7.70e-03	-0.13	3.51e-01	2.27e-02	-2.70e-01	8.93e-02	4.41e-02
XYKI	5.89e-02	-0.04	-1.36e-02	4.25e-02	-7.70e-02	-2.78e-01	7.24e-02

Classical Machine Learning Assignment¶

1. Is the wheat in this dataset homogeneous, random or does it fall into distinct types?¶

2. If the wheat is homogenenous or random, then identify the overall characteristics of the complete sample.¶

3. If there are distinct types of wheat in the dataset, then identify the characteristics of each type of wheat.¶

4. What is the relationship between the input variables (first 40 features) and the 5 response variables?¶

5. Are there specific features which ‘drive’ specific responses?¶

a) If so, how?¶

b) What are the important features and what is their relative impact?¶

6. If there are different types of wheat present in the dataset, then which type of wheat has the greatest impact on each response variable?¶

Libraries Used¶

References¶